Peri-implant marginal bone resorption is a common threat to long-term implant stability and success of implantation. Abnormally pathological MBL without effective treatment is often accompanied with peri‐implant mucositis.[17] The previous research on MBL mainly concentrated on treatment, but the cause of MBL and the way to predict it has rarely been studied. Factors such as trabecular bone microstructure parameters that possibly affect MBL during bone remolding were known to us, but the role of trabecular bone in this progression was still unclear. In this study, we analyzed morphological parameters of trabecular bone in patients with MBL during bone remolding by CBCT scan and CT Analyzer analysis. Our results demonstrated the differences and correlations of morphological parameters between the controls and abnormal MBL cases during bone remolding.
Trabecular bone is a defining factor for quality of bones structured with a trabecular pattern.[18] Trabecular microarchitecture can be displayed by several morphological parameters, including BV/TV, i.S, Tb.Pf, SMI, Tb.Th and Tb.N. In the present study, during the early stage of functional loading, CBCT analysis exhibited a significant increase of SMI and Tb.Pf in abnormal cases, while BV/TV, i.S, and Tb.N increased in normal controls. This finding revealed that severe MBL was accompanied with inferior trabecular microarchitecture at the early post-loading stage, while normal MBL showed an increased bone mass. A previous study reported a similar trend: the preservation and improvement of trabecular microarchitecture always brought about a better therapeutic benefit for osteoporosis at multiple skeletal sites.[19] Our result indicated SMI, Tb.Pf, BV/TV, i.S, and Tb.N at the early stage could be predictors for MBL level in the later stage, which was inconsistent with the previous studies in which bone remolding was believed to be closely associated with implant prognosis.[20]
Inspired by the above findings, we analyzed all variables using correlation and covariance matrices. All results relevant to morphological variables were confirmed with a significant difference and reasonable collinearity. Among them, SMI, Tb.Pf, Tb.N, BS/BV, and BV/TV manifested a higher correlation with MBL, while other morphological variables could not bring a noteworthy contribution. These findings revealed that SMI and Tb.Pf were the best determinants of the MBL level. Both of them could reflect the structure quality of the trabecular bone and gained a decline in peri-implant of abnormal cases. However, Tb.N and BV/TV in normal controls were superior to other morphological variables, which accorded with the previous research.[20]
Accurate prediction of abnormally severe MBL at an early stage can aid in seeking possible factors that can influence normal bone remolding and prevent implant failure. Previous studies attempted to evaluate the risk of MBL using the proportion of cancellous bone,[11] crown-to-implant ratio,[21] bone texture and cortical width,[22] respectively. However, a single predictive factor may insufficient to accurately predict MBL occurrence, because MBL is a multifactorial outcome. As mentioned before, various factors, including cortical bone thickness, smoking, periodontitis, SMI, Tb.Pf, and BV/TV, might be regarded together as a complex to predict MBL.
Utilizing machine learning algorithms to predict the occurrence of MBL, the current study has obtained satisfying manifestation. An accurate prediction can warn patients and clinicians of possible factors that impact bone remolding, so as to intervene and avoid MBL timely.
A crucial finding that we extracted from the overall results is that the abnormal rise of SMI is tightly connected with severe MBL. Under proper stress, both osteocytes and osteoblasts, as effector cells of mechanical stimulation, can enhance bone formation and strength via a series of complex regulations.[23, 24] After bone remolding, the microdamage and repairment of the trabecular bone may typically achieve a dynamical balance.[25] However, the inferior trabecular structure model reveals that this balance is broken. If we find the detrimental factor that threatens the balance at an early stage and take intervention, severe MBL and implant failure may possibly be prevented, which is also the original intention of our research.
This study was restricted to patients receiving implant treatment in mandible. Mandible has thicker alveolar crest cortical bone than the maxilla.[26] Meanwhile, the maxilla implant is commonly in conjunction with bone augmentation. Previous predictive machine learning models had similar sample sizes in disease prediction.[16, 27, 28] Adding implant occlusion-related variables, such as overbite, overjet, median line location, or molar relationship, may enable better performing models based on the factors related to stress concentration. The current models all achieved considerable accuracy without incorporating the above variables, which was interesting. Due to the limited number of applicable subjects, we entitled this study as a preliminary one.